Evaluation of Potting Possibilities for RF Circuits up

to 24 GHz
Magnus O. Olbrich, Wolfgang Kiermeier, Erwin Biebl
Technische Universitaet Muenchen , Fachgebiet Hoechstfrequenztechnik, Arcisstr. 21, 80333 Muenchen, Germany,
+498928923930, [email protected]

Electrical parameters of potting materials are usually specified dielectric ε in its complex form becomes relevant the same
by the manufacturer only up to some MHz. We have determined holds true for the potting material.
the real part of relative permittivity and loss tangent of a potting
compound at discrete frequencies up to 24 GHz by use of ring The difference for the relative permittivity lies in the
resonators. The results are validated by the design and susceptibility of electromagnetic fields. Potting compounds are
measurement of a bandpass filter at 11 GHz. For designing this adhesives or resins. For understanding rf characteristics the
test circuit the gained potting material parameters were already molecule structure has to be considered. In general two
included in the simulation. different molecular structures exist. Nonpolar ones like
polyethylene have symmetric molecule structures and therefore
Keywords-component; potting; encapsulation; are only susceptible for electromagnetic fields by the
displacement of the electron cloud compared to the core.
I. INTRODUCTION Whereas polar ones like polyvinyl chloride (PVC) show a
dipole moment to electromagnetic field by their asymmetric
Encapsulants and potting compounds provide
structure. These two configurations are shown exemplarily in
environmental protection against moisture or mechanical stress
Fig. 1. For rf applications only nonpolar materials are suitable
and electrical isolation for electronic components. Their use is
due to their much lesser losses.
widely spread for dc and low frequency ac circuits, but not yet
for rf circuits due to their influence on the rf circuit
characteristics.
However also for rf circuits potting would be interesting
from a mechanical and moisture protection point of view. The
manufacturers of potting materials specify the rf relevant
material properties usually only up to some MHz. The very
ideal potting material characteristics would be the same as from
air. Therewith a circuit can be designed and measured with air,
and afterwards be produced and potted without any change. A
potting material with the same characteristics as the substrate
would result in a different kind of line. E.g. a microstrip line
could be transferred into a stripline. Therefore the circuit has to
be potted also for measurement. But this would only be a
mentionable effort for very low lots of the final product. For a
mass market product this additional effort is negligible and
good line characteristics, e.g. real TEM propagation without Figure 1. Example of nonpolar and polar molecules.
dispersion instead of quasi TEM mode would be achieved.
Especially at high frequencies dielectric constants are often
References [1] and [2] discuss microstrip characteristics determined by use of resonators. For rf design the relative
with a dielectric overlay in general. This work evaluates the permittivity in its complex form has to be known. We specify
possibilities of potting planar rf circuits up to 24 GHz in terms the real part of the relative permittivity εr’ and the tanδ,
of additional attenuation and inclusion of the gained material because these parameters are given to the simulation tool and
parameters into the simulation. are equipollent to the complex form.

II. BASICS From the various forms of resonators the ring resonator in a
microstrip line technology was chosen. Compared to simple
Whereas for dc potting parameters like heat conductivity or straight open ended designs, ring resonators have the best
dielectric strength play the most important role, for higher quality factor. This is because of the elimination of open end
frequencies additional parameters have to be taken into fringing and thereby sharp defined resonant length.
account. As for the substrate material the permittivity of the Furthermore, a considerably large contact area to the potting
resin is realized this way. The ring itself as well as the feeding
lines were designed for a line impedance of 50 Ohms at the
respective designated frequency. Coupling was realized at both
sides as shown in Fig. 2 using 0.15 mm wide gaps. This causes
some distortion in the results due to the influence of the potting
material not only on the ring but also on the coupling. However
the test design measurements showed that this effect is
negligible, because the 2D analysis simulation is not only
applied to the resonator itself, but in fact to the whole circuit
including the coupling section. This can easily be achieved by a
simulation with a complete upper hemisphere consisting of the
respective potting material.
The final dielectric constants cannot be derived directly Figure 3. Mounting of the 12GHz & 24 GHz resonator.
from the comparison of empty and potted circuit measurements
because of the unknown asymmetric electric field distribution
over and below the microstrip. Therefore the measured
resonator center frequencies and quality factors have to be
transferred into a 2D EM simulation program, varying the IV. MEASUREMENTS
simulation dielectric values of the overlay to achieve One of the evaluated potting materials (amongst some
coincidence of measured and simulated parameters. epoxy and polyurethane resins) was the KC-Produkte
Preliminary tests and simulations showed that thin layers of SH 33650 that is originally intended to cushion vibrations and
potting compounds were sufficient for satisfactory precision. to firm up the mechanical strength. It is a copolymer of
No further frequency shift or damping occurred above a height polyethylene and polyvinyl acetate (EVA) that was specified
of approximately 5 mm. Thus all measurements were taken by KC at 50 Hz (εr’=2.1; tanδ<0.001). In contrast to the more
using 10 mm high layers to certainly eliminate thickness traditional 2-component based potting systems it has to be
influence. applied to the circuit by heating it up to 120 °C where it
becomes liquid.

Figure 2. Ring resonator with side coupling.

Figure 4. Potted resonator for 5.8 GHz.

III. CIRCUITS
The frequency range was evaluated by certain discrete Fig. 4 shows the potted 5.8 GHz resonator that was measured
values to keep the effort manageable. Resonator circuits were by a network vector analyzer. In this work |S21| was measured
designed and fabricated for 2.4 GHz, 5.8 GHz, 12 GHz and and used for the following analysis. The potting compound
24 GHz, because these frequencies represent the needs of most changes the resonators centre frequency and the quality factor
technical applications. To keep the size of the circuit small for compared to a measurement before the potting. The gained
2.4 GHz the substrate CER-10 with a high permittivity of 9.5 parameters for the different frequencies are given in table I. It
was used. For the other circuits TLP-5-0100 substrate with a is obvious that they are comparable to those of usual
permittivity of 2.2 was utilized. The high permittivity substrate microwave substrates.
would shrink these circuits to a size that brings the fabrication
accuracy to its limits. Moreover, and especially important for TABLE I. MEASURED PARAMETERS FOR KC SH 33650
the higher frequencies, the TLP substrate has lower losses.
Different mounting and supports were used to comply the Frequency 2.4 GHz 5.8 GHz 12 GHz 24 GHz
diverse demands of the frequency bands. Exemplarily the
circuit used for 12 GHz, and also for 24 GHz by exciting the ε r’ 2.25 2.27 2.27 2.15
first harmonic, is shown in Fig. 3. A small paper tube on top of 0.001 0.007 0.01 0.01
tanδ
resonator and coupling serves as a resin container and prevents
the mounting from being glued. At 2.4 GHz and 5.8 GHz
standard SMA printed-circuit-connectors in conjunction with a
metal resin-container were applied, see Fig. 4.
 parameters fc=11.00 GHz, A=-1.22 dB and Q=11.3 in the
A further series of measurements shall demonstrate the simulation. The measurement results correspond very good the
comparably poor rf performance of a fast curing 2-component predicted simulation results. The measured parameters are
epoxy resin, that is extensively used for more traditional low fc=11.00 GHz, A=-1.92 dB and Q=12. The marginal higher
frequency potting applications, namely the R+G “5 minute attenuation A also exists in the measurement without potting
epoxy”. The values for its relative permittivity and loss tangent and is originated by the measurement test set. Fig. 6 shows the
can be found in table II (measured 15 min after potting) and simulated S21 in magnitude and Fig. 7 the measured S21
table III (measured 1 day after potting). The much worse magnitude of the potted 11 GHz bandpass filter.
values of the uncured / partially cured epoxy are due to ionic
Figure 6. Simulation of the potted 11 GHz bandpass.
components in the resin.

TABLE II. MEASURED PARAMETERS FOR EPOXY RESIN, 15 MIN CURED

Frequency 2.4 GHz 5.8 GHz 12 GHz 24 GHz

ε r’ 3.75 3.55 3.25 2.90

tanδ 0.09 0.09 0.08 0.1

TABLE III. MEASURED PARAMETERS FOR EPOXY RESIN, 1 DAY CURED

Frequency 2.4 GHz 5.8 GHz 12 GHz 24 GHz Figure 7. Measurment results of the 11 GHz bandpass.

ε r’ 3.5 3.4 3.0 2.8

tanδ 0.03 0.07 0.06 0.08

V. TEST CIRCUIT VI. CONCLUSION

By use of ring resonators the relevant electrical parameters
of several potting compounds for rf application were
determined at certain frequencies up to 24 GHz. These
parameters are the real part of the relative permittivity and the
loss tangent. To validate the gained parameters a coupled line
bandpass filter at 11 GHz was designed including the potting
compound parameters already in the simulation. The
measurement results of this test circuit corresponds very good
with its simulation.
Figure 5. Original (left) and potted 11 GHz bandpass.
REFERENCES
To prove the applicability of the gained material [1] I. J. Bahl and S. Stuchly, “Analysis of a microstrip covered with a lossy
parameters a test circuit was designed. A coupled line bandpass dielectric,” IEEE Trans. on MTT, vol. 28, no. 2, pp. 104–109, February
1980.
filter with a center frequency of 11 GHz was used. The gained
[2] K. K. Joshi, R. D. Pollard and V. Postoyalko, “Microstrip with dielectric
KC material parameters were applied to the simulation tool overlay: variational analysis and validation,” IEE Proc. Microw.
ADS and the circuit design included the potting layer from its Antennas Propag., vol. 141, no. 2, pp. 138–140, April 1994.
very beginning. The original and the potted circuit are shown .
in Fig. 5. The left circuit represents the original given bandpass
filter designed for traditional operation in air, while the right
circuit shows the redesigned potted filter with identical center
frequency and quality factor. The circuit showed the